MM-StanceDet: Retrieval-Augmented Multi-modal Multi-agent Stance Detection

2026-04-30

AI Summary

This work addresses key challenges in multimodal stance detection—such as difficulties in image-text fusion, lack of contextual grounding, cross-modal ambiguity, and fragility of single-pass reasoning—by proposing the first framework that integrates retrieval augmentation, multi-agent collaboration, and structured reasoning. The approach leverages retrieval-augmented context anchoring to enrich input representations, employs specialized multimodal analysis agents for fine-grained interpretation, and incorporates debate and self-reflection mechanisms to enhance reasoning robustness. Extensive experiments across five benchmark datasets demonstrate that the proposed architecture substantially outperforms current state-of-the-art methods, thereby validating the efficacy of multi-agent coordination and structured reasoning in tackling complex multimodal stance detection tasks.

Abstract

Multimodal Stance Detection (MSD) is crucial for understanding public discourse, yet effectively fusing text and image, especially with conflicting signals, remains challenging. Existing methods often face difficulties with contextual grounding, cross-modal interpretation ambiguity, and single-pass reasoning fragility. To address these, we propose Retrieval-Augmented Multi-modal Multi-agent Stance Detection (MM-StanceDet), a novel multi-agent framework integrating Retrieval Augmentation for contextual grounding, specialized Multimodal Analysis agents for nuanced interpretation, a Reasoning-Enhanced Debate stage for exploring perspectives, and Self-Reflection for robust adjudication. Extensive experiments on five datasets demonstrate MM-StanceDet significantly outperforms state-of-the-art baselines, validating the efficacy of its multi-agent architecture and structured reasoning stages in addressing complex multimodal stance challenges.

Problem

• Multimodal Stance Detection
• cross-modal interpretation
• contextual grounding
• conflicting signals
• stance detection

Innovation

• Retrieval Augmentation
• Multi-agent Framework
• Multimodal Stance Detection
• Reasoning-Enhanced Debate
• Self-Reflection